Genome-Wide Expression Profiling of Human Lymphoblastoid Cell Lines

ORIGINAL ARTICLE Genome-wide expression proﬁling of human lymphoblastoid cell lines implicates integrin beta-3 in the mode of action of antidepressants

K Oved1,2,6, A Morag1,3,6, M Pasmanik-Chor4, M Rehavi3, N Shomron2,5 and D Gurwitz1

Selective serotonin reuptake inhibitors (SSRIs) are the first-line treatment for major depression. However, the link between inhibition of serotonin reuptake and remission from depression remains controversial: in spite of the rapid onset of serotonin reuptake inhibition, remission from depression takes several weeks, presumably reflecting synaptogenesis/neurogenesis and neuronal rewiring. We compared genome-wide expression profiles of human lymphoblastoid cell lines from unrelated individuals following treatment with 1 mM paroxetine for 21 days with untreated control cells and examined which genes and microRNAs (miRNAs) showed the most profound and consistent expression changes. ITGB3, coding for integrin beta-3, showed the most consistent altered expression (1.92-fold increase, P ¼ 7.5 Â 10 À 8) following chronic paroxetine exposure. Using genome-wide miRNA arrays, we observed a corresponding decrease in the expression of two miRNAs, miR-221 and miR-222, both predicted to target ITGB3. ITGB3 is crucial for the activity of the serotonin transporter (SERT), the drug target of SSRIs. Moreover, it is presumably required for the neuronal guidance activity of CHL1, whose expression was formerly identified as a tentative SSRI response biomarker. Further genes whose expression was significantly modulated by chronic paroxetine are also implicated in neurogenesis. Surprisingly, the expression of SERT or serotonin receptors was not modified. Our findings implicate ITGB3 in the mode of action of SSRI antidepressants and provide a novel link between CHL1 and the SERT. Our observations suggest that SSRIs may relieve depression primarily by promoting neuronal synaptogenesis/neurogenesis rather than by modulating serotonin neurotransmission per se.

Translational Psychiatry (2013) 3, e313; doi:10.1038/tp.2013.86; published online 15 October 2013 Keywords: CD61; GPIIIa; human lymphoblastoid cell lines; ITGB3; microRNAs; SSRI antidepressants; serotonin transporter

INTRODUCTION however, human Positron emission tomography imaging studies Selective serotonin reuptake inhibitors (SSRIs) have been estab- yielded ambiguous findings.16–18 lished as the first-line treatment for major depression since their We set out to study this question by using the unbiased introduction in the early 1980s.1–3 These drugs are known to bind genome-wide approach, rather than examining changes in with high affinity and inhibit the activity of the presynaptic expression levels of presumed relevant genes. We chose to study serotonin transporter (SERT), the specific SSRIs drug target. The human cells for the genome-wide search, rather than using animal biogenic amine hypothesis of depression, albeit controversial, models for depression, which have some limitations.19,20 argues that SSRI drugs relieve depression primarily via their SERT Moreover, we compared human cell lines from four unrelated blocking action that in turn increases the central nervous system donors for improved power: we profiled the expression of genes serotonin availability, in particular in the thalamo–prefrontal and microRNAs (miRNAs) of human lymphoblastoid cell lines 4,5 cortex pathway. However, a key unresolved issue is that, (LCLs) chronically exposed to 1 mM paroxetine. LCLs have been although maximal SERT inhibition is achieved within few days of instrumental in pharmacogenomic discovery.21–23 We recently starting SSRI drug medication, it takes around 4 weeks for the utilized LCLs from unrelated healthy individuals for conducting a onset of remission of depression.1–3 Recovery from depres- genome-wide transcriptomic microarray based search for SSRI sion was proposed to require synaptogenesis as well as neuro- sensitivity biomarkers and reported several genes as tentative genesis6–9 combined with migration of hippocampal neuronal SSRI response biomarkers.24,25 Among these, CHL1, coding for a stem cells,7,10–12 which may account for this long delay. However, cell adhesion protein implicated in neurogenesis and synapto- the mechanisms by which SSRIs-mediated SERT blockade leads to genesis,26–28 seems a promising biomarker, as it is essential for the brain synaptogenesis/neurogenesis or neuronal rewiring correct neuronal wiring from the thalamus to the prefrontal remain enigmatic. Animal studies have shown downregulation cortex,29 a neuronal pathway crucial for mood control. Moreover, of the brain SERT expression by chronic SSRI medication;13–15 CHL1 was reported as a potential SSRI sensitivity biomarker in a

1Department of Human Molecular Genetics and Biochemistry, Sackler Faculty of Medicine, Tel-Aviv University, Tel-Aviv, Israel; 2Department of Cell and Developmental Biology, Sackler Faculty of Medicine, Tel-Aviv University, Tel-Aviv, Israel; 3Department of Physiology and Pharmacology, Sackler Faculty of Medicine, Tel-Aviv University, Tel-Aviv, Israel; 4Bioinformatics Unit, George Wise Faculty of Life Sciences, Tel-Aviv University, Tel-Aviv, Israel and 5Sagol School of Neuroscience, Tel-Aviv University, Tel-Aviv, Israel. Correspondence: Dr N Shomron, Department of Cell and Developmental Biology, Sackler Faculty of Medicine, Tel-Aviv University, Tel-Aviv 69978, Israel or Dr D Gurwitz, Department of Human Molecular Genetics and Biochemistry, Sackler Faculty of Medicine, Tel-Aviv University, Tel-Aviv 69978, Israel. E-mail: [email protected] or [email protected] 6These authors contributed equally to this work. Received 4 June 2013; revised 15 August 2013; accepted 8 September 2013 ITGB3 is implicated in the mode of action of SSRI antidepressants K Oved et al 2 large study in major depression patients.30 The identification of CHL1 as a potential SSRI response biomarker was further suppor- Gene Forward primer Reverse primer ted by our genome-wide miRNA expression studies, where levels of miR-151-3p, which targets CHL1, were related to SSRI GUSB (control) CTGCTGGCTACTACTTGAAGATG GAGTTGCTCACAAAGGTCAC 25 ITGB3 ACCAGTAACCTGCGGATTG CAGGTGGTCTTCATATCATAGC responsiveness of LCLs. HECW2 GTTACAGGCACATCCAGCATT ATGTATGCGCTCTGGGAA Here we describe our genome-wide transcriptomic findings MAL ACATAGCCACTCTGCTC CTGGGTTTTCAGCTCAAGTTC from microarray expression profiling experiments in human LCLs KLHL24 GCAGTCTGTGCTCTAAGGAA ACGGCTGTTGATTCTTC chronically treated with the SSRI paroxetine. We report that the expression of ITGB3 (integrin beta-3) as well as miR-221 and miR- 222, which target ITGB3, exhibited the most consistent expression level changes following chronic paroxetine exposure in LCSs miRNA Mature miRNA sequence ABI assay ID representing four unrelated male individuals. Our genome-wide expression profiling observations highlight the importance of U6 snRNA GTGCTCGCTTCGGCAGCACATATACTAAAATTGGAA 1973 (control) CGATACAGAGAAGATTAGCATGGCCCCTGCGCAAG neurogenesis and synaptogenesis in the mode of action of SSRI GATGACACGCAAATTCGTGAAGCGTTCCATATTTT antidepressant drugs and further support a key role for CHL1 in miR-221 AGCUACAUUGUCUGCUGGGUUUC 524 determining SSRI sensitivity. These findings further point to a key miR-222 AGCUACAUCUGGCUACUGGGU 2276 role of cell adhesion proteins such as CHL1 and ITGB3 in remission from depression.

MATERIALS AND METHODS RESULTS Human LCLs and chronic paroxetine treatment Cultured human LCLs from four unrelated adult male donors Human LCLs were obtained from the National Laboratory for the Genetics (Ashkenazi Jewish ancestry) growing under optimal conditions (cell 24,25 5 of Israeli Populations (NLGIP) at Tel-Aviv University as described. The density of 3 to 8 Â 10 cells per ml) were exposed to 1 mM paroxetine cell lines were immortalized from the peripheral blood lymphocytes for 21 days, whereas parallel cultures of the same LCLs received of healthy adult male donors of Ashkenazi Jewish ancestry. Four NLGIP similar volumes of phosphate-buffered saline. Following RNA cell lines were used, coded 1126, 1131, 1235 and 1371. Cells were main- extraction, Affymetrix GeneChip Human Gene 1.0 ST arrays or tained in Roswell Park Memorial Institute (RPMI) medium supplemented 1 GeneChip miRNA 2.0 arrays were used for detecting the expression with 10% fetal bovine serum and antibiotics (100 U ml À penicillin; 100 mgmlÀ 1 streptomycin) and kept at a temperature of 37 1C, with 6% levels of genes and miRNAs, respectively, according to the manufacturer’s protocol (see Methods). Data were analyzed by CO2 and 100% humidity. Paroxetine was purchased from Sigma-Aldrich (St Louis, MO, USA) and solutions were prepared in phosphate-buffered Partek Genomics Suite and normalized as described in Materials and saline. For chronic treatment, cell lines in logarithmic growth were exposed methods. Tables 1 and 2 present genes and miRNAs, respectively, to 1 mM paroxetine for 21 days. Fresh paroxetine (from a 1000-fold stock whose expression levels showed 41.5-fold (41.4-fold for miRNAs) solution) was added on each feeding the cell cultures according to added difference and Po0.001 in preparations from LCL cultures chronically medium volume (every 2 to 3 days). Control cultures (grown in parallel) exposed to paroxetine compared with parallel controls. As shown, received similar volume of phosphate-buffered saline on each feeding. ITGB3 (coding for ITGB3; also known as platelet glycoprotein IIIa and CD61) exhibited the most statistically significant change in expres- RNA extraction sion levels following 21 days paroxetine exposure, (1.925-fold À 8 Total RNA purification was achieved using phenol-chloroform extraction25; increased expression; P ¼ 7.50 Â 10 ) for the four LCLs. Four genes cells were centrifuged and then lysed using Tri-reagent (T9424, Sigma- (MAL, HECW2, ITGB3 and KLHL24) and two miRNAs (miR-221 and miR- Aldrich), followed by RNA separation using chloroform and precipitation 222) from Tables 1 and 2 were selected for validation with real-time using isopropanol. RNA quality was checked using RNAse free, 1% agarose PCR in each of the four individual LCLs (Figures 1 and 2), confirming gel and was quantified using a NanoDrop spectrophotometer (ND-1000). the effects of paroxetine on their expression levels. The choice of The spectrophotometric absorbance parameters of the samples were: 260/ genes and miRNAs was made as they are known to be expressed in 280 nm 41.8 and 260/230 nm 42.0. the brain tissues and implicated in neurogenesis. As shown in Figures 1 and 2, the altered expression levels of these selected genes and Microarray experiments miRNAs following chronic paroxetine exposure were closely similar in RNAs and miRNAs were compared for each of the four human LCLs the LCLs of four unrelated donors. between chronic paroxetine exposure and controls. Affymetrix GeneChip Human Gene 1.0 ST arrays and Affymetrix GeneChip miRNA 2.0 arrays were used for gene and miRNA expression analysis, respectively, according to DISCUSSION the instruction manuals (Affymetrix, Santa Clara, CA, USA). Microarray Transcriptomic changes following chronic paroxetine exposure analysis was performed on CEL files using Partek Genomics Suite TM The expression levels of 14 genes were changed by 41.5-fold and (Partek, St Louis, MO, USA). Data were normalized and summarized with P 0.001 following chronic paroxetine exposure in the four tested the robust multi-average method.31 Batch effect removal was applied for o the different samples, to remove individual variations, followed by one- human LCLs from unrelated male donors (Table 1). Among them ITGB3 exhibited the most statistically significant change: its way analysis of variance. Genes and miRNAs of interest that were -8 differentially expressed when comparing paroxetine-treated LCLs and expression increased on average by 1.92-fold (P ¼ 7.50 Â 10 )as controls (Po0.001 and fold difference cutoff 1.5 (1.4-fold for miRNAs)) confirmed by real-time PCR (Table 1, Figure 1). This observation is were obtained. unexpected, as none of these 14 genes are related to established serotonin signaling or metabolism. ITGB3 has neither been previously implicated in the etiology of major depression nor in Real-time PCR the mode of action of SSRI antidepressant drugs. Of note, this Real-time PCR assays were performed for validating the microarray data for identification is supported by our genome-wide array based four selected genes and two selected miRNAs using the same RNA preparations, as described.24,25 Comparative critical threshold (Ct) values, identification of decreased expression levels of two miRNAs, miR- obtained by real-time PCR analysis, were used for relative quantification of 221 and miR-222, both predicted to target ITGB3,andasconfirmed genes or miRNAs expression and determination of fold-change of by real-time PCR, using the same LCLs and experimental protocol expression. Individual forward and reverse primers were used as (Table 2, Figure 2). The targeting of ITGB3 by both miR-221 and miR- described below: 222 is predicted by each of the following software: TargetScan,

Translational Psychiatry (2013), 1 – 6 & 2013 Macmillan Publishers Limited ITGB3 is implicated in the mode of action of SSRI antidepressants K Oved et al 3 Moreover, paroxetine was employed in our recently published Table 1. Genes whose expression was affected by chronic paroxetine LCLs studies on tentative SSRI response biomarkers,24,25 so that exposure of LCLs combining our previous and current studies using the same drug Gene Fold-changea P value allows us to integrate the data and to propose a new model for SSRI antidepressants mode of action (below). ITGB3 1.925 7.50 Â 10 À 8 The identification of elevated ITGB3 expression following HECW2 1.812 4.69 Â 10 À 6 chronic paroxetine exposure is intriguing. Integrins, including OVOS À 1.517 7.31 Â 10 À 6 ITGB3, are implicated in cell adhesion and connectivity and are À 5 CD109 1.586 1.21 Â 10 essential for synaptogenesis. For example, ITGB3 regulates RNF144A 1.550 4.81 Â 10 À 5 32 À 5 excitatory synaptic strength and hippocampal AMPA receptors MAL 2.316 5.55 Â 10 expression.33 Remission from depression is presumed to be KLHL24 1.50 7.89 Â 10 À 5 À 4 dependent upon establishing new neuronal connections, in TSPAN12 1.54 1.77 Â 10 6–11 CCL28 1.51 2.21 Â 10 À 4 particular in the prefrontal cortex. CD180 À 1.51 2.52 Â 10 À 4 MAL (T-cell differentiation protein), whose expression was also PLOD2 1.71 5.52 Â 10 À 4 increased by chronic paroxetine in the same LCLs (2.316-fold; CD244 1.87 8.77 Â 10 À 4 P ¼ 5.55 Â 10 À 5; Table 1, Figure 1) is implicated in myelin form- DTX1 1.50 9.36 Â 10 À 4 ation. Notably, its expression was reduced in post-mortem temporal À 4 DNASE1L3 1.96 9.44 Â 10 cortex of major depression patients.34 Other genes listed in Table 1 Abbreviation: LCLs, lymphoblastoid cell lines. The 14 listed genes exhibited were not implicated in major depression to our knowledge. Notably, 41.5-fold difference and Po0.001 in expression levels in four LCLs TSPAN12 (tetraspanin 12; Table 1) was implicated in cleavage of the a 35 following treatment for 21 days with 1 mM paroxetine. Fold-change amyloid precursor protein and its deletion was reported in represents expression levels following paroxetine exposure compared autism,36 a disorder associated with incorrect brain circuitry. In with controls grown and studied by microarrays in parallel. As shown, 12 of addition, DTX1 (deltex homolog 1; Table 1) was implicated in these 14 top genes were upregulated by chronic paroxetine exposure. neurogenesis37 and in gliogenic specification of mouse Expression levels were determined with whole-genome expression mesencephalic neural crest cells.38 Lastly, PLOD2 (procollagen- microarrays and genes are arranged by increasing P-values. The expression lysine, 2-oxoglutarate 5-dioxygenase 2; Table 1) was detected in a differences for four selected genes (in bold font) were confirmed by real- genome-wide search for mouse genes implicated in axonal time PCR experiments (Figure 1). Note: two transcripts with no assigned 39 genes (LOC100289612; C1orf186) were excluded. regeneration. Together, our genome-wide expression profiling observations in paroxetine-treated LCLs point to SSRI-mediated upregulation of genes implicated in synaptogenesis/neurogenesis, supporting the concept that SSRI drugs relieve depression by promoting these processes.6–11 Table 2. MicroRNAs whose expression was affected by chronic Among the miRNAs identified by our microarray based paroxetine exposure of LCLs genome-wide transcriptomic profiling study miR-221 and miR- MiRNA Fold-changea P value 222 exhibited 1.525- and 1.416-fold lower expression, respectively, following chronic paroxetine exposure of the same LCLs (Table 2). miR-3195 1.527 8.90 Â 10 À 8 These findings were confirmed by real-time PCR (Figure 2). As miR-1246 2.406 7.01 Â 10 À 6 both these miRNAs are predicted to target ITGB3 with conserved miR-221 À 1.525 7.22 Â 10 À 6 binding sites (Supplementary Table S1), the observations on their À 5 miR-1290 4.001 1.42 Â 10 reduced expression complement the observed higher expression miR-1263 À 1.411 8.36 Â 10 À 5 À 5 levels of ITGB3 following chronic paroxetine. Of note, a genome- miR-550-star À 1.428 8.63 Â 10 wide miRNA microarray study found that miR-221 was down- miR-29c-star 1.640 2.68 Â 10 À 4 miR-3178 1.440 3.95 Â 10 À 4 regulated in the hippocampi of rats chronically treated with À 4 lithium, a mood-stabilizer drug also employed for augmenting miR-222 À 1.416 5.07 Â 10 40 miR-664 À 1.615 5.19 Â 10 À 4 antidepressant therapy. Abbreviations: LCLs, lymphoblastoid cell lines; miRNA, micro RNA. The 10 listed miRNAs exhibited 41.4-fold difference and Po0.001 in expression Strengths and constraints levels in four LCLs following treatment for 21 days with 1 mM paroxetine. aFold-change represents expression levels following paroxetine exposure An obvious strength of our study is the hypothesis-free nature of compared with controls grown and studied by microarrays in parallel. As genome-wide expression studies. Moreover, we examined gene and shown, one half of these top miRNAs were upregulated by chronic miRNA expression in LCLs from four unrelated male donors, so that paroxetine exposure. Expression levels were determined with whole- confounders caused by the presence of unique polymorphic DNA genome expression microarrays and miRNAs are arranged by increasing sequence alleles or epigenomic modifications are unlikely to have P-values. The expression differences for two selected miRNAs (in bold font) contributed to our observations. Nevertheless, care is called for were confirmed by real-time PCR experiments (Figure 2). According to when interpreting genome-wide data and our observations should PubMed search (27 May 2013) the only miRNAs in this list reportedly thus be considered as tentative until independently confirmed. Our expressed in the human brain are miR-221 and miR-222. study has several additional limitations as discussed below. First, we utilized LCLs from healthy unrelated donors for studying the transcriptional effects of chronic paroxetine expo- DIANA, TargetRank and PITA (Supplementary Table S1; see ref. 25 for sure. Although these immortalized cell lines capture the natural details about these miRNA target prediction tools). variation of the human genome and epigenome,21–23 one has to We chose the period of 21 days for the genome-wide micro- keep in mind that transcriptomic drug effects may differ in the array experiments based on preliminary real-time experiments brain compared with the blood-born cells. In particular, compared (not shown) and as this period reflects the onset of recovery from with neurons, LCLs exhibit low methylation signatures41 so that depressive symptoms in SSRI-medicated patients.1,2 Paroxetine epigenomic effects on transcription may not faithfully represent was selected for these experiments as it has the highest affinity for the situation in neurons. Yet, Zhang et al.42 have proposed that, SERT, the SSRI drug target, among approved SSRI drugs. This although the methylome and the transcriptome are modified choice minimizes the probability for offtarget drug effects. during the establishment of LCLs, the cells maintain a relatively

& 2013 Macmillan Publishers Limited Translational Psychiatry (2013), 1 – 6 ITGB3 is implicated in the mode of action of SSRI antidepressants K Oved et al 4

Figure 1. Expression changes for MAL, HECW2, ITGB3 and KLHL24 following chronic paroxetine exposure. Data are shown for microarrays (a, b) and real-time PCR (c, d) experiments as averages for four lymphoblastoid cell lines (LCLs) (a, c) or for each individual cell line (b, d), respectively. See Table 1 and Materials and Methods for experimental details. Note the close similarity for the altered gene expression in LCLs representing four unrelated donors.

Figure 2. Expression changes for miR-221 and miR-222 following chronic paroxetine exposure. Data are shown for microarrays (a, b) and real-time PCR (c, d) experiments as averages for four lymphoblastoid cell lines (LCLs) (a, c) or for each individual cell line (b, d), respectively. See Table 2 and Materials and Methods for experimental details.

stable status during cell culturing. Furthermore, the alterations are do not express CYP2D6, the major phase-I drug metabolizing enzyme not random events, so LCLs generated separately from the same of paroxetine,44 hence the observed transcriptomic changes were donor will have more similar methylation and transcription unlikely mediated by a paroxetine metabolite. patterns than those generated from different donors. Thus, the Lastly, the statistical significance of the changes of genes and different transcriptomic profiles of LCLs may in part reflect miRNAs expression levels is low for a genome-wide study, different donors’ methylation status. reflecting our small sample size (eight microarrays, two for each Second, we utilized a concentration of paroxetine (1 mM), which was of four human LCLs). Indeed, only the expression of one gene, about three- to fivefold higher than its therapeutic plasma ITGB3, exhibited genome-wide significance. However, genome- concentrations in paroxetine-treated major depression patients.43 wide expression profiling arrays are less likely to generate false This concentration was chosen for the microarray experiments based hits compared with single nucleotide polymorphism arrays which on preliminary real-time PCR findings (not shown). Moreover, we did typically contain nearly one million single nucleotide polymorph- nottestthestabilityofparoxetinewith our experimental protocol and ism probes. In addition, we selected four genes and two miRNAs its concentrations might have been reduced by binding to albumin in for validation by real-time PCR experiments and found good our serum-containing medium (however, fresh paroxetine was added agreement with our array data. Moreover, two of the miRNAs most whenever feeding the cells, see Materials and Methods). Of note, LCLs notably affected by chronic paroxetine exposure, miR-221 and

Translational Psychiatry (2013), 1 – 6 & 2013 Macmillan Publishers Limited ITGB3 is implicated in the mode of action of SSRI antidepressants K Oved et al 5 miR-222, showed decreased expression levels corresponding with the observed upregulation of their target gene ITGB3. These observations support the novelty of the increased expression of ITGB3 following chronic paroxetine (further examples for inverse correlations of miRNAs levels with their target genes are shown in Supplementary Table S1). Considering these experimental design aspects, our observations should be viewed as tentative until validated by further studies on the transcriptomic effects of SSRI drugs. The implication of the genes and miRNAs reported here for the etiology and treatment of major depression should ideally be examined with the blood samples of large cohorts of major depression patients, both before and following several weeks of treatment with SSRI drugs. Comparing such transcriptomic changes between good and poor SSRI responders may contribute to the personalized treatment of major depression.

A hypothetical potential role for ITGB3 in mediating the action of SSRI antidepressants Figure 3. Hypothetical model depicting the cell membrane proteins We have previously reported that low basal expression levels of encoded by CHL1 and SLC6A4 (SERT) competing on a limited pool of CHL1 (close homologue of L1), which codes for a cell adhesion integrin beta-3 (ITGB3) protein. (a) At low CHL1 expression levels, protein implicated in neurogenesis and synaptogenesis, correlated more ITGB3 is available for supporting serotonin transporter (SERT) serotonin-uptake activity, hence higher sensitivity to SSRI drugs is with higher sensitivity of LCLs to growth inhibition by SSRI observed. (b) At high CHL1 expression levels (and similar SERT and drugs.24 We subsequently reported that mir-151-3p, predicted to 25 ITGB3 expression levels as depicted in panel a) more ITGB3 interacts target CHL1, exhibited higher expression levels in the same LCLs, with CHL1 and less ITGB3 is available for supporting SERT serotonin- thereby lending support for a role for CHL1 levels in SSRI uptake activity, hence lower sensitivity to SSRI drugs is observed. sensitivity. It was therefore surprising that the expression levels of either CHL1 or mir-151-3p were not modified by chronic paroxetine exposure in our current study. In addition, none of the dynamic action of CHL1 for supporting synaptogenesis/neurogen- genes whose expression levels were modified by chronic esis. As such events are presumably crucial for successful paroxetine exposure (Table 1) code for the SERT, serotonin remission from depression,6–12 our working hypothesis supports receptors or serotonin metabolizing enzymes, all of which have a role for ITGB3 in the treatment of depression. Moreover, our received attention in the context of research on the mode of findings and model, suggestive of a tentative role for elevated action of antidepressant drugs. expression of cell adhesion proteins for remission from We postulate a tentative working hypothesis (Figure 3) for the depression, may in part explain the enigmatic slow onset of the involvement of ITGB3, whose expression was most consistently antidepressant action of SSRI drugs. Elucidating the roles of ITGB3, increased following chronic paroxetine, in the mode of action of CHL1 and additional cell adhesion proteins in the mode of action SSRIs and its relation to the previously reported role of CHL1 of SSRI antidepressants requires studies with animal models for expression levels in modulating SSRI sensitivity of LCLs.24 Our depression and eventually clinical studies. working hypothesis is built upon the observation that the integrin beta-3 subunit encoded by ITGB3 is required for the activity of the SERT, the drug target of SSRIs, as evident from the drastically NOTE ADDED IN PROOF reduced serotonin-uptake activity in ITGB3-knockout mice.45 A new study51 demonstrates that expression of ITGB3 is essential Notably, these mice exhibit absence of preference for social for serotonin transporter activity in mouse brain. novelty and increased grooming in novel environments, behaviors relevant for autism spectrum disorder.46 Neuroanatomical assessment of these mice indicated that many brain regions had CONFLICT OF INTEREST significantly different relative volumes, including a smaller corpus The authors declare no conflict of interest. callosum volume and bilateral decreases in the hippocampus, striatum and cerebellum, all relevant to autism.47 Together these findings suggest that ITGB3 is crucial for correct neuroanatomical ACKNOWLEDGMENTS 29 development of the brain, a property it shares with CHL1. This study was supported by the Chief Scientist office, Ministry of Health, Israel, in the Integrins also interact with CHL1 and are required for its cell frame of ERA-Net Neuron. N Shomron is supported by the Israel Cancer Research migration and neurite outgrowth promoting action48–50; albeit, Fund (ICRF); Wolfson Family Charitable Fund; Earlier.org—Friends for an Earlier Breast the exact integrin subtype(s) implicated in CHL1-mediated neurite Cancer Test; Claire and Amedee Maratier Institute for the Study of Blindness and outgrowth await identification. We therefore postulate that CHL1 Visual Disorders; I-CORE Program of the Planning and Budgeting Committee, Israel. D competes with SERT on a limited cell membrane reservoir of Gurwitz is supported by the Yoran Institute for Human Genome Research at Tel Aviv ITGB3. Low levels of CHL1 allow more ITGB3 to interact with SERT University. We thank the anonymous donors of the NLGIP biobank at Tel Aviv University, Israel, whose altruism and trust in biomedical research have made this and support its action; thus, lower CHL1 expression correlate with study possible. higher SERT activity, hence higher SSRI responsiveness, manifested as more robust inhibition of LCLs growth by SSRIs in 24,25 our previous studies. Along the lines of this model (Figure 3), REFERENCES cells exposed to chronic SSRI compensate for the blocked SERT by 1 Gelenberg AJ. A review of the current guidelines for depression treatment. J Clin upregulating the expression of its co-activator ITGB3 (rather than Psychiatry 2010; 71:e15. the expression of SERT itself). Extrapolating our hypothesis from 2 Rayner L, Price A, Evans A, Valsraj K, Hotopf M, Higginson IJ. Antidepressants for LCLs to the brain serotonergic neurons, the resulting higher ITGB3 the treatment of depression in palliative care: systematic review and meta-ana- levels that follow chronic SSRI treatment in turn allow more lysis. Palliat Med 2011; 25: 36–51.

& 2013 Macmillan Publishers Limited Translational Psychiatry (2013), 1 – 6 ITGB3 is implicated in the mode of action of SSRI antidepressants K Oved et al 6 3 Cipriani A, Purgato M, Furukawa TA, Trespidi C, Imperadore G, Signoretti A et al. 30 Clark SL, Adkins DE, Aberg K, Hettema JM, McClay JL, Souza RP, van den Oord EJ. Citalopram versus other anti-depressive agents for depression. Cochrane Data- Pharmacogenomic study of side-effects for antidepressant treatment options in base Syst Rev 2012; 7: CD006534. STAR*D. Psychol Med 2012; 42: 1151–1162. 4 van Praag HM, Kahn R, Asnis GM, Lemus CZ, Brown SL. Therapeutic indications for 31 Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U et al. serotonin-potentiating compounds: a hypothesis. Biol Psychiatry 1987; 22: 205–212. Exploration, normalization, and summaries of high density oligonucleotide array 5 Price LH, Charney DS, Delgado PL, Heninger GR. Lithium and serotonin function: probe level data. Biostatistics 2003; 4: 249–264. implications for the serotonin hypothesis of depression. Psychopharmacology 32 Cingolani LA, Goda Y.. Differential involvement of beta3 integrin in pre- and (Berl) 1990; 100: 3–12. postsynaptic forms of adaptation to chronic activity deprivation. Neuron Glia Biol 6 Hanson ND, Owens MJ, Nemeroff CB. Depression, antidepressants, and neuro- 2008; 4: 179–187. genesis: a critical reappraisal. Neuropsychopharmacology 2011; 36: 2589–60. 33 Pozo K, Cingolani LA, Bassani S, Laurent F, Passafaro M, Goda Y. b3 integrin interacts 7 Eisch AJ, Petrik D. Depression and hippocampal neurogenesis: a road to remis- directly with GluA2 AMPA receptor subunit and regulates AMPA receptor expression? Science 2012; 338: 72–75. sion in hippocampal neurons. Proc Natl Acad Sci USA 2012; 109: 1323–1328. 8 Bambico FR, Belzung C.. Novel insights into depression and antidepressants: a 34 Aston C, Jiang L, Sokolov BP. Transcriptional profiling reveals evidence for sig- synergy between synaptogenesis and neurogenesis? Curr Top Behav Neurosci naling and oligodendroglial abnormalities in the temporal cortex from patients 2012; 15: 243–291. with major depressive disorder. Mol Psychiatry 2005; 10: 309–322. 9 Mateus-Pinheiro A, Pinto L, Bessa JM, Morais M, Alves ND, Monteiro S, Patrıćio P, 35 Xu D, Sharma C, Hemler ME. Tetraspanin12 regulates ADAM10-dependent clea- Almeida OF, Sousa N. Sustained remission from depressive-like behavior depends vage of amyloid precursor protein. FASEB J 2009; 23: 3674–3681. on hippocampal neurogenesis. Transl Psychiatry 2013; 3: e210. 36 Okamoto N, Hatsukawa Y, Shimojima K, Yamamoto T. Submicroscopic deletion in 10 Thomas RM, Peterson DA. Even neural stem cells get the blues: evidence for a 7q31 encompassing CADPS2 and TSPAN12 in a child with autism spectrum dis- molecular link between modulation of adult neurogenesis and depression. Gene order and PHPV. Am J Med Genet A 2011; 155A: 1568–1573. Expr 2008; 14: 183–193. 37 Kishi N, Tang Z, Maeda Y, Hirai A, Mo R, Ito M et al. Murine homologs of 11 Eyre H, Baune BT. Neuroplastic changes in depression: a role for the immune deltex define a novel gene family involved in vertebrate Notch signaling and system. Psychoneuroendocrinology 2012; 37: 1397–1416. neurogenesis. Int J Dev Neurosci 2001; 19: 21–35. 12 Danzer SC. Depression, stress, epilepsy and adult neurogenesis. Exp Neurol. 2012; 38 Fujita K, Yasui S, Shinohara T, Ito K. Interaction between NF-kB signaling and 233: 22–32. Notch signaling in gliogenesis of mouse mesencephalic neural crest cells. 13 Benmansour S, Cecchi M, Morilak DA, Gerhardt GA, Javors MA, Gould GG, Frazer A. Mech Dev 2011; 128: 496–509. Effects of chronic antidepressant treatments on serotonin transporter function, 39 Read ML, Mir S, Spice R, Seabright RJ, Suggate EL, Ahmed Z et al. Profiling RNA density, and mRNA level. J Neurosci 1999; 19: 10494–10501. interference (RNAi)-mediated toxicity in neural cultures for effective short inter- 14 Benmansour S, Owens WA, Cecchi M, Morilak DA, Frazer A. Serotonin clearance fering RNA design. J Gene Med 2009; 11: 523–534. in vivo is altered to a greater extent by antidepressant-induced downregulation of 40 Zhou R, Yuan P, Wang Y, Hunsberger JG, Elkahloun A, Wei Y et al. Evidence for the serotonin transporter than by acute blockade of this transporter. J Neurosci selective microRNAs and their effectors as common long-term targets for the 2002; 22: 6766–6772. actions of mood stabilizers. Neuropsychopharmacology 2009; 34: 1395–1405. 15 Mirza NR, Nielsen EØ, Troelsen KB. Serotonin transporter density and anxiolytic- 41 Horike S, Cai S, Miyano M, Cheng JF, Kohwi-Shigematsu T. Loss of silent-chromatin like effects of antidepressants in mice. Prog Neuropsychopharmacol Biol Psychiatry looping and impaired imprinting of DLX5 in Rett syndrome. Nat Genet 2005; 37: 2007; 31: 858–866. 31–40. 16 Meyer JH. Imaging the serotonin transporter during major depressive disorder 42 Zhang Z, Liu J, Kaur M, Krantz ID. Characterization of DNA methylation and its and antidepressant treatment. J Psychiatry Neurosci 2007; 32: 86–102. association with other biological systems in lymphoblastoid cell lines. Genomics 17 Huang Y, Zheng MQ, Gerdes JM. Development of effective PET and SPECT ima- 2012; 99: 209–219. ging agents for the serotonin transporter: has a twenty-year journey reached its 43 Yasui-Furukori N, Nakagami T, Kaneda A, Inoue Y, Suzuki A, Otani K, Kaneko S. destination? Curr Top Med Chem 2010; 10: 1499–1526. Inverse correlation between clinical response to paroxetine and plasma drug 18 Descarries L, Riad M. Effects of the antidepressant fluoxetine on the subcellular concentration in patients with major depressive disorders. Hum Psychopharmacol localization of 5-HT1A receptors and SERT. Philos Trans R Soc Lond B Biol Sci 2012; 2011; 26: 602–608. 367: 2416–2425. 44 Vincent M, Oved K, Morag A, Pasmanik-Chor M, Oron-Karni V, Shomron N, Gurwitz 19 Overstreet DH. Modeling depression in animal models. Methods Mol Biol 2012; D. Genome-wide transcriptomic variations of human lymphoblastoid cell lines: 829: 125–144. insights from pairwise gene-expression correlations. Pharmacogenomics 2012; 13: 20 Harro J. Animal models of depression vulnerability. Curr Top Behav Neurosci 2013; 1893–1904. 14: 29–54. 45 Carneiro AM, Cook EH, Murphy DL, Blakely RD. Interactions between integrin 21 Sie L, Loong S, Tan EK.. Utility of lymphoblastoid cell lines. J Neurosci Res 2009; 87: alphaIIbbeta3 and the serotonin transporter regulate serotonin transport and 1953–1959. platelet aggregation in mice and humans. J Clin Invest 2008; 118: 1544–1552. 22 Shim SM, Nam HY, Lee JE, Kim JW, Han BG, Jeon JP. MicroRNAs in human lym- 46 Carter MD, Shah CR, Muller CL, Crawley JN, Carneiro AM, Veenstra-VanderWeele J. phoblastoid cell lines. Crit Rev Eukaryot Gene Expr 2012; 22: 189–196. Absence of preference for social novelty and increased grooming in integrin b3 23 Wheeler HE, Dolan ME. Lymphoblastoid cell lines in pharmacogenomic discovery knockout mice: initial studies and future directions. Autism Res 2011; 4: 57–67. and clinical translation. Pharmacogenomics 2012; 13: 55–70. 47 Ellegood J, Henkelman RM, Lerch JP. Neuroanatomical assessment of the integrin 24 Morag A, Pasmanik-Chor M, Oron-Karni V, Rehavi M, Stingl JC, Gurwitz D. Genome-wide b3 mouse model related to autism and the serotonin system using high resolu- expression profiling of human lymphoblastoid cell lines identifies CHL1 as a putative tion MRI. Front Psychiatry 2012; 3:37. SSRI antidepressant response biomarker. Pharmacogenomics 2011; 12: 171–184. 48 Buhusi M, Midkiff BR, Gates AM, Richter M, Schachner M, Maness PF. Close 25 Oved K, Morag A, Pasmanik-Chor M, Oron-Karni V, Shomron N, Rehavi M, homolog of L1 is an enhancer of integrin-mediated cell migration. J Biol Chem Stingl JC, Gurwitz D. Genome-wide miRNA expression profiling of human 2003; 278: 25024–25031. lymphoblastoid cell lines identifies tentative SSRI antidepressant response 49 Demyanenko GP, Schachner M, Anton E, Schmid R, Feng G, Sanes J, Maness PF. biomarkers. Pharmacogenomics 2012; 13: 1129–1139. Close homolog of L1 modulates area-specific neuronal positioning and dendrite 26 Ango F, Wu C, Van der Want JJ, Wu P, Schachner M, Huang ZJ. Bergmann glia and orientation in the cerebral cortex. Neuron 2004; 44: 423–437. the recognition molecule CHL1 organize GABAergic axons and direct innervation 50 Schlatter MC, Buhusi M, Wright AG, Maness PF. CHL1 promotes Sema3A-induced of Purkinje cell dendrites. PLoS Biol 2008; 6: e103. growth cone collapse and neurite elaboration through a motif required for recruit- 27 Ye H, Tan YL, Ponniah S, Takeda Y, Wang SQ, Schachner M et al. Neural recog- ment of ERM proteins to the plasma membrane. J Neurochem 2008; 104:731–744. nition molecules CHL1 and NB-3 regulate apical dendrite orientation in the 51 Whyte A, Jessen T, Varney S, Carneiro AM. Serotonin transporter and integrin beta neocortex via PTP alpha. EMBO J 2008; 27: 188–200. 3 genes interact to modulate serotonin uptake in mouse brain. Neurochem Int. 28 Tian N, Leshchyns’ka I, Welch JH, Diakowski W, Yang H, Schachner M, Sytnyk V. Lipid advance online publication, 28 September 2013; doi:pii:S0197-0186(13)00241-6. raft-dependent endocytosis of close homolog of adhesion molecule L1 (CHL1) 10.1016/j.neuint.2013.09.014 (e-pub ahead of print). promotes neuritogenesis. JBiolChem2012; 287: 44447–44463. 29 Demyanenko GP, Siesser PF, Wright AG, Brennaman LH, Bartsch U, Schachner M, This work is licensed under a Creative Commons Attribution- Maness PF. L1 and CHL1 cooperate in thalamocortical axon targeting. Cereb NonCommercial-NoDerivs 3.0 Unported License. To view a copy of Cortex 2011; 21: 401–412. this license, visit http://creativecommons.org/licenses/by-nc-nd/3.0/

Supplementary Information accompanies the paper on the Translational Psychiatry website (http://www.nature.com/tp)

Translational Psychiatry (2013), 1 – 6 & 2013 Macmillan Publishers Limited